A Novel Edge Feature Correlation Algorithm for Real-Time Computer Vision-Based Molten Weld Pool Measurements

The design, testing, and implementation of a versatile algorithm developed to provide robust real-time processing of images for use with a visionbased computer control system is discussed. A generic approach to the implementation of the image-processing algorithm has been employed, allowing a wide variety of imaging devices to be used. The particular application of this work involves the implementation of a weld face or upper surface sensor for GTAW process measurement and automation based upon image edge feature detection. The objective is to utilize the measurements generated by the processing algorithm within a feedback-based process control system. This control system is intended to provide active adjustment of the key process control parameters in order to compensate for deviations from a predefined set of reference or "base line" conditions that may occur in practice. This provides an attractive and potentially more efficient alternative to more traditional "open loop" type control schemes that would normally be employed. The combination of this application and the algorithm that has been developed represent a novel use for computer-based image processing.